AI-based malware detection in IoT networks within smart cities: A survey

Abstract

The exponential expansion of Internet of Things (IoT) applications in smart cities has significantly pushed smart city development forward. Intelligent applications have the potential to enhance systems' efficiency, service quality, and overall performance. Smart cities, intelligent transportation networks, and other influential infrastructure are the main targets of cyberattacks. These attacks have the potential to undercut the security of important government, commercial, and personal information, placing privacy and confidentiality at risk. Multiple scientific studies indicate that Smart City cyberattacks can result in millions of euros in financial losses due to data compromise and loss. The importance of anomaly detection rests in its ability to identify and analyze illegitimacy within IoT data. Unprotected, infected, or suspicious devices may be unsafe for intrusion attacks, which have the potential to enter several machines within a network. This interferes with the network's provision of customer service in terms of privacy and safety. The objective of this study is to assess procedures for detecting malware in the IoT using artificial intelligence (AI) approaches. To identify and prevent threats and malicious programs, current methodologies use AI algorithms such as support vector machines, decision trees, and deep neural networks. We explore existing studies that propose several methods to address malware in IoT using AI approaches. Finally, the survey highlights current issues in this context, including the accuracy of detection and the cost of security concerns in terms of detection performance and energy consumption.